Re-basing the core a government runs on

The Government's core network had become a constraint on the whole public service. Frequent failures were degrading citizen services, and the digital transformation agenda could not proceed on an ageing, unstable foundation.

This was never a hardware refresh. I treated it as re-basing the whole-of-government technology foundation: review the entire system, chart a roadmap, choose a target, and put lasting design standards and governance in place. Reliability was the floor. A durable foundation was the goal.

The most consequential call on the project was one nobody had asked me to make. The favoured design reused the same platform family as the ageing core, on grounds of familiarity. I caught that the platform line was heading to end-of-life well inside the Government's own roadmap, re-opened the decision, and moved the design to a platform built for the future service profile.

A wrong platform commitment is an organisational cost, not a technical one. Had the original choice gone in, a whole-of-government core would have been stranded on unsupported hardware precisely as demand peaked. The platform chosen instead remains well within its capacity headroom today.

Because these services tolerate no sustained outage, the migration had to happen on a live network inside maintenance windows of four to five hours. I designed a staged switchover that kept the old and new cores running side by side and moved one department, ministry, site or subnet at a time.

Any single move could be checked and, if needed, reversed within its window. Risk was contained to one unit at a time, never the whole estate.

The environment also had no security standards in place, so I adopted the NIST Cybersecurity Framework and CIS Controls ahead of any government mandate: baseline device hardening, least-privilege access control, and segmentation by criticality.

The new core went live across the entire government estate and met every acceptance criterion. Latency, throughput and availability all improved, and so did the things people feel: application launch times, transaction speed, day-to-day productivity.

The stronger evidence came later. The architecture standards I documented were adopted as the baseline for subsequent government technology projects. A design that outlives its designer is the real test of enterprise architecture, and this one passed it.